We enrolled 479 patients with subarachnoid hemorrhage in a multicenter, randomized, double-blind, placebo-controlled trial to determine whether treatment with the antifibrinolytic agent tranexamic acid improves outcome by preventing rebleeding. At three months there was no statistical difference between the outcomes in the tranexamic acid group and the control group. Of the 173 patients who died, 84 had received tranexamic acid and 89 placebo (95 per cent confidence interval for the difference in mortality rate, -6 to 11 per cent). Similarly, when analysis was restricted to patients with an angiographically demonstrated aneurysm, there was no significant difference between the groups. This absence of effect was not due to a lack of antifibrinolytic action, since the rate of rebleeding was reduced from 24 per cent in the control group to 9 per cent in the tranexamic acid-treated group (chi-square = 18.07, P less than 0.001), but resulted from a concurrent increase in the incidence of ischemic complications (15 per cent in the control group and 24 per cent in the tranexamic acid group; chi-square = 8.07, P less than 0.01). We conclude that until some method can be found to minimize ischemic complications, tranexamic acid is of no benefit in patients with subarachnoid hemorrhage.